In present communication, a solar dryer having semi-transparent photovoltaic thermal (SPVT) roof has been used to dry bitter gourd. Experimentation has been done and results were validated with linear regression method through mathematical modelling as well as for non-linear Artificial Neural Network (ANN) modelling has been used. The experimental data namely, ambient air temperature, moisture evaporated, bitter gourd surface temperatures, solar intensity on different surfaces of dryer, room temperature and relative humidity have been recorded for the climatic condition of Varanasi, UP, India. The numerical value of convective heat transfer coefficients for bitter gourd found to be between 0.69–14.45 W/m2 K. The results for mass transfer (moisture evaporated) during drying of bitter gourd from theoretical model (linear regression method and ANN) and experiments showed good agreement. Further, net electrical energy, thermal energy and overall thermal energy were found to be 0.62, 3.77 and 5.41 kWh/m2/day respectively. The payback period at 10\% interest rate and 4\% inflation rate with ₹ 50 ($0.77) and ₹ 100 ($1.54) saving rate per kg were found to be 1.47 and 0.72 years respectively. © 2020, Springer-Verlag GmbH Germany, part of Springer Nature.